dc.description.abstract | Previous pharmacogenetics studies of 5-fluorouracil (5-FU) have focused on coding variants, and only four of these variants showed consistent association, which explained a small fraction of the heritability of 5-FU-associated toxicity. No previous studies have systematically investigated non-coding variants that regulate expression of 5-FU catabolic genes and their roles in 5-FU-associated toxicity in cancer patients. This dissertation developed approaches that integrated multiple functional genomics datasets to identify variants predicting expression of genes in normal tissues and prioritize additional potential regulatory variants for 5-FU catabolic genes including DPYD, DPYS and UPB1. To investigate the role of these genetic variants, a cohort of 424 colorectal cancer patients was established using resources from the Vanderbilt cancer registry and the biobank, BioVU. It was found that a decrease in the genetically predicted expression of catabolic genes was associated with an increased risk of severe toxicity (OR per one standard deviation decrease = 1.28, 95% CI: 1.05-1.59, P = 0.02), and the genetic risk score combining all prioritized potential regulatory variants were positively associated with the risk of severe toxicity (OR per effect allele = 1.33, 95% CI: 1.09-1.61, P = 0.004). In combination with known genetic and non-genetic risk factors, these genetic markers together provided moderate discriminatory accuracy for predicting severe toxicity. This dissertation has supported the important role of regulatory variants in the risk of 5-FU associated toxicity, identified a panel of markers of potential clinical use and provided a new direction for future pharmacogenetics studies. | |